chm_lib.c

It is a chm file viewer lib with source code

/* $Id: chm_lib.c,v 1.19 2003/09/07 13:01:43 jedwin Exp $ */
/***************************************************************************
 *             chm_lib.c - CHM archive manipulation routines               *
 *                           -------------------                           *
 *                                                                         *
 *  author:     Jed Wing <[e-mail]>                         *
 *  version:    0.3                                                        *
 *  notes:      These routines are meant for the manipulation of microsoft *
 *              .chm (compiled html help) files, but may likely be used    *
 *              for the manipulation of any ITSS archive, if ever ITSS     *
 *              archives are used for any other purpose.                   *
 *                                                                         *
 *              Note also that the section names are statically handled.   *
 *              To be entirely correct, the section names should be read   *
 *              from the section names meta-file, and then the various     *
 *              content sections and the "transforms" to apply to the data *
 *              they contain should be inferred from the section name and  *
 *              the meta-files referenced using that name; however, all of *
 *              the files I've been able to get my hands on appear to have *
 *              only two sections: Uncompressed and MSCompressed.          *
 *              Additionally, the ITSS.DLL file included with Windows does *
 *              not appear to handle any different transforms than the     *
 *              simple LZX-transform.  Furthermore, the list of transforms *
 *              to apply is broken, in that only half the required space   *
 *              is allocated for the list.  (It appears as though the      *
 *              space is allocated for ASCII strings, but the strings are  *
 *              written as unicode.  As a result, only the first half of   *
 *              the string appears.)  So this is probably not too big of   *
 *              a deal, at least until CHM v4 (MS .lit files), which also  *
 *              incorporate encryption, of some description.               *
 *                                                                         *
 * switches:    CHM_MT:        compile library with thread-safety          *
 *                                                                         *
 * switches (Linux only):                                                  *
 *              CHM_USE_PREAD: compile library to use pread instead of     *
 *                             lseek/read                                  *
 *              CHM_USE_IO64:  compile library to support full 64-bit I/O  *
 *                             as is needed to properly deal with the      *
 *                             64-bit file offsets.                        *
 ***************************************************************************/

/***************************************************************************
 *                                                                         *
 *   This program is free software; you can redistribute it and/or modify  *
 *   it under the terms of the GNU Lesser General Public License as        *
 *   published by the Free Software Foundation; either version 2.1 of the  *
 *   License, or (at your option) any later version.                       *
 *                                                                         *
 ***************************************************************************/

#include "chm_lib.h"

#ifdef CHM_MT
#define _REENTRANT
#endif

#include "lzx.h"

#include <stdlib.h>
#include <string.h>
#ifdef CHM_DEBUG
#include <stdio.h>
#endif

#if __sun || __sgi
#include <strings.h>
#endif

/* RWE 6/13/2003 */
#ifdef _WIN32_WCE
#define FREEBUF(x) free(x)
#else
#define FREEBUF(x) /* do nothing */
#endif

#ifdef WIN32
#include <windows.h>
#include <malloc.h>
#ifdef _WIN32_WCE
#define strcasecmp _stricmp
#define strncasecmp _strnicmp
#else
#define strcasecmp stricmp
#define strncasecmp strnicmp
#endif
#else
/* basic Linux system includes */
#define _XOPEN_SOURCE 500
#include <unistd.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
/* #include <dmalloc.h> */
#endif

/* includes/defines for threading, if using them */
#ifdef CHM_MT
#ifdef WIN32
#define CHM_ACQUIRE_LOCK(a) do {                        \
        EnterCriticalSection(&(a));                     \
    } while(0)
#define CHM_RELEASE_LOCK(a) do {                        \
        EnterCriticalSection(&(a));                     \
    } while(0)

#else
#include <pthread.h>

#define CHM_ACQUIRE_LOCK(a) do {                        \
        pthread_mutex_lock(&(a));                       \
    } while(0)
#define CHM_RELEASE_LOCK(a) do {                        \
        pthread_mutex_unlock(&(a));                     \
    } while(0)

#endif
#else
#define CHM_ACQUIRE_LOCK(a) /* do nothing */
#define CHM_RELEASE_LOCK(a) /* do nothing */
#endif

#ifdef WIN32
#define CHM_NULL_FD (INVALID_HANDLE_VALUE)
#define CHM_USE_WIN32IO 1
#define CHM_CLOSE_FILE(fd) CloseHandle((fd))
#else
#define CHM_NULL_FD (-1)
#define CHM_CLOSE_FILE(fd) close((fd))
#endif

/*
 * defines related to tuning
 */
#ifndef CHM_MAX_BLOCKS_CACHED
#define CHM_MAX_BLOCKS_CACHED 5
#endif

/*
 * architecture specific defines
 *
 * Note: as soon as C99 is more widespread, the below defines should
 * probably just use the C99 sized-int types.
 *
 * The following settings will probably work for many platforms.  The sizes
 * don't have to be exactly correct, but the types must accommodate at least as
 * many bits as they specify.
 */

/* i386, 32-bit, Windows */
#ifdef WIN32
typedef unsigned char           UChar;
typedef __int16                 Int16;
typedef unsigned __int16        UInt16;
typedef __int32                 Int32;
typedef unsigned __int32        UInt32;
typedef __int64                 Int64;
typedef unsigned __int64        UInt64;

/* I386, 32-bit, non-Windows */
/* Sparc        */
/* MIPS         */
/* PPC          */
#elif __i386__ || __sun || __sgi || __ppc__
typedef unsigned char           UChar;
typedef short                   Int16;
typedef unsigned short          UInt16;
typedef long                    Int32;
typedef unsigned long           UInt32;
typedef long long               Int64;
typedef unsigned long long      UInt64;

/* x86-64 */
/* Note that these may be appropriate for other 64-bit machines. */
#elif __x86_64__
typedef unsigned char           UChar;
typedef short                   Int16;
typedef unsigned short          UInt16;
typedef int                     Int32;
typedef unsigned int            UInt32;
typedef long                    Int64;
typedef unsigned long           UInt64;

#else

/* yielding an error is preferable to yielding incorrect behavior */
#error "Please define the sized types for your platform in chm_lib.c"
#endif

/* GCC */
#ifdef __GNUC__
#define memcmp __builtin_memcmp
#define memcpy __builtin_memcpy
#define strlen __builtin_strlen

#elif defined(WIN32)
static int ffs(unsigned int val)
{
    int bit=1, idx=1;
    while (bit != 0  &&  (val & bit) == 0)
    {
        bit <<= 1;
        ++idx;
    }
    if (bit == 0)
        return 0;
    else
        return idx;
}

#endif

/* utilities for unmarshalling data */
static int _unmarshal_char_array(unsigned char **pData,
                                 unsigned int *pLenRemain,
                                 char *dest,
                                 int count)
{
    if (count <= 0  ||  (unsigned int)count > *pLenRemain)
        return 0;
    memcpy(dest, (*pData), count);
    *pData += count;
    *pLenRemain -= count;
    return 1;
}

static int _unmarshal_uchar_array(unsigned char **pData,
                                  unsigned int *pLenRemain,
                                  unsigned char *dest,
                                  int count)
{
        if (count <= 0  ||  (unsigned int)count > *pLenRemain)
        return 0;
    memcpy(dest, (*pData), count);
    *pData += count;
    *pLenRemain -= count;
    return 1;
}

static int _unmarshal_int16(unsigned char **pData,
                            unsigned int *pLenRemain,
                            Int16 *dest)
{
    if (2 > *pLenRemain)
        return 0;
    *dest = (*pData)[0] | (*pData)[1]<<8;
    *pData += 2;
    *pLenRemain -= 2;
    return 1;
}

static int _unmarshal_uint16(unsigned char **pData,
                             unsigned int *pLenRemain,
                             UInt16 *dest)
{
    if (2 > *pLenRemain)
        return 0;
    *dest = (*pData)[0] | (*pData)[1]<<8;
    *pData += 2;
    *pLenRemain -= 2;
    return 1;
}

static int _unmarshal_int32(unsigned char **pData,
                            unsigned int *pLenRemain,
                            Int32 *dest)
{
    if (4 > *pLenRemain)
        return 0;
    *dest = (*pData)[0] | (*pData)[1]<<8 | (*pData)[2]<<16 | (*pData)[3]<<24;
    *pData += 4;
    *pLenRemain -= 4;
    return 1;
}

static int _unmarshal_uint32(unsigned char **pData,
                             unsigned int *pLenRemain,
                             UInt32 *dest)
{
    if (4 > *pLenRemain)
        return 0;
    *dest = (*pData)[0] | (*pData)[1]<<8 | (*pData)[2]<<16 | (*pData)[3]<<24;
    *pData += 4;
    *pLenRemain -= 4;
    return 1;
}

static int _unmarshal_int64(unsigned char **pData,
                            unsigned int *pLenRemain,
                            Int64 *dest)
{
    Int64 temp;
    int i;
    if (8 > *pLenRemain)
        return 0;
    temp=0;
    for(i=8; i>0; i--)
    {
        temp <<= 8;
        temp |= (*pData)[i-1];
    }
    *dest = temp;
    *pData += 8;
    *pLenRemain -= 8;
    return 1;
}

static int _unmarshal_uint64(unsigned char **pData,
                             unsigned int *pLenRemain,
                             UInt64 *dest)
{
    UInt64 temp;
    int i;
    if (8 > *pLenRemain)
        return 0;
    temp=0;
    for(i=8; i>0; i--)
    {
        temp <<= 8;
        temp |= (*pData)[i-1];
    }
    *dest = temp;
    *pData += 8;
    *pLenRemain -= 8;
    return 1;
}

static int _unmarshal_uuid(unsigned char **pData,
                           unsigned int *pDataLen,
                           unsigned char *dest)
{
    return _unmarshal_uchar_array(pData, pDataLen, dest, 16);
}

/* names of sections essential to decompression */
static const char _CHMU_RESET_TABLE[] =
        "::DataSpace/Storage/MSCompressed/Transform/"
        "{7FC28940-9D31-11D0-9B27-00A0C91E9C7C}/"
        "InstanceData/ResetTable";
static const char _CHMU_LZXC_CONTROLDATA[] =
        "::DataSpace/Storage/MSCompressed/ControlData";
static const char _CHMU_CONTENT[] =
        "::DataSpace/Storage/MSCompressed/Content";
static const char _CHMU_SPANINFO[] =
        "::DataSpace/Storage/MSCompressed/SpanInfo";

/*
 * structures local to this module
 */

/* structure of ITSF headers */
#define _CHM_ITSF_V2_LEN (0x58)
#define _CHM_ITSF_V3_LEN (0x60)
struct chmItsfHeader
{
    char        signature[4];           /*  0 (ITSF) */
    Int32       version;                /*  4 */
    Int32       header_len;             /*  8 */
    Int32       unknown_000c;           /*  c */
    UInt32      last_modified;          /* 10 */
    UInt32      lang_id;                /* 14 */
    UChar       dir_uuid[16];           /* 18 */
    UChar       stream_uuid[16];        /* 28 */
    UInt64      unknown_offset;         /* 38 */
    UInt64      unknown_len;            /* 40 */
    UInt64      dir_offset;             /* 48 */
    UInt64      dir_len;                /* 50 */
    UInt64      data_offset;            /* 58 (Not present before V3) */
}; /* __attribute__ ((aligned (1))); */

static int _unmarshal_itsf_header(unsigned char **pData,
                                  unsigned int *pDataLen,
                                  struct chmItsfHeader *dest)
{
    /* we only know how to deal with the 0x58 and 0x60 byte structures */
    if (*pDataLen != _CHM_ITSF_V2_LEN  &&  *pDataLen != _CHM_ITSF_V3_LEN)
        return 0;

    /* unmarshal common fields */
    _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
    _unmarshal_int32     (pData, pDataLen, &dest->version);
    _unmarshal_int32     (pData, pDataLen, &dest->header_len);
    _unmarshal_int32     (pData, pDataLen, &dest->unknown_000c);
    _unmarshal_uint32    (pData, pDataLen, &dest->last_modified);
    _unmarshal_uint32    (pData, pDataLen, &dest->lang_id);
    _unmarshal_uuid      (pData, pDataLen,  dest->dir_uuid);
    _unmarshal_uuid      (pData, pDataLen,  dest->stream_uuid);
    _unmarshal_uint64    (pData, pDataLen, &dest->unknown_offset);
    _unmarshal_uint64    (pData, pDataLen, &dest->unknown_len);
    _unmarshal_uint64    (pData, pDataLen, &dest->dir_offset);
    _unmarshal_uint64    (pData, pDataLen, &dest->dir_len);

    /* error check the data */
    /* XXX: should also check UUIDs, probably, though with a version 3 file,
     * current MS tools do not seem to use them.
     */
    if (memcmp(dest->signature, "ITSF", 4) != 0)
        return 0;
    if (dest->version == 2)
    {
        if (dest->header_len < _CHM_ITSF_V2_LEN)
            return 0;
    }
    else if (dest->version == 3)
    {
        if (dest->header_len < _CHM_ITSF_V3_LEN)
            return 0;
    }
    else
        return 0;

    /* now, if we have a V3 structure, unmarshal the rest.
     * otherwise, compute it
     */
    if (dest->version == 3)
    {
        if (*pDataLen != 0)
            _unmarshal_uint64(pData, pDataLen, &dest->data_offset);
        else
            return 0;
    }
    else
        dest->data_offset = dest->dir_offset + dest->dir_len;

    return 1;
}

/* structure of ITSP headers */
#define _CHM_ITSP_V1_LEN (0x54)
struct chmItspHeader
{
    char        signature[4];           /*  0 (ITSP) */
    Int32       version;                /*  4 */
    Int32       header_len;             /*  8 */
    Int32       unknown_000c;           /*  c */
    UInt32      block_len;              /* 10 */
    Int32       blockidx_intvl;         /* 14 */
    Int32       index_depth;            /* 18 */
    Int32       index_root;             /* 1c */
    Int32       index_head;             /* 20 */
    Int32       unknown_0024;           /* 24 */
    UInt32      num_blocks;             /* 28 */
    Int32       unknown_002c;           /* 2c */
    UInt32      lang_id;                /* 30 */
    UChar       system_uuid[16];        /* 34 */
    UChar       unknown_0044[16];       /* 44 */
}; /* __attribute__ ((aligned (1))); */

static int _unmarshal_itsp_header(unsigned char **pData,
                                  unsigned int *pDataLen,
                                  struct chmItspHeader *dest)
{
    /* we only know how to deal with a 0x54 byte structures */
    if (*pDataLen != _CHM_ITSP_V1_LEN)
        return 0;

    /* unmarshal fields */
    _unmarshal_char_array(pData, pDataLen,  dest->signature, 4);
    _unmarshal_int32     (pData, pDataLen, &dest->version);
    _unmarshal_int32     (pData, pDataLen, &dest->header_len);
    _unmarshal_int32     (pData, pDataLen, &dest->unknown_000c);
    _unmarshal_uint32    (pData, pDataLen, &dest->block_len);
    _unmarshal_int32     (pData, pDataLen, &dest->blockidx_intvl);
    _unmarshal_int32     (pData, pDataLen, &dest->index_depth);
    _unmarshal_int32     (pData, pDataLen, &dest->index_root);
    _unmarshal_int32     (pData, pDataLen, &dest->index_head);
    _unmarshal_int32     (pData, pDataLen, &dest->unknown_0024);
    _unmarshal_uint32    (pData, pDataLen, &dest->num_blocks);
    _unmarshal_int32     (pData, pDataLen, &dest->unknown_002c);
    _unmarshal_uint32    (pData, pDataLen, &dest->lang_id);
    _unmarshal_uuid      (pData, pDataLen,  dest->system_uuid);
    _unmarshal_uchar_array(pData, pDataLen, dest->unknown_0044, 16);

    /* error check the data */
    if (memcmp(dest->signature, "ITSP", 4) != 0)
        return 0;
    if (dest->version != 1)
        return 0;
    if (dest->header_len != _CHM_ITSP_V1_LEN)
        return 0;

    return 1;
}